CLINICAL RESEARCH www.jasn.org Clinical Features and Outcomes of Immune Checkpoint Inhibitor–Associated AKI: A Multicenter Study Frank B. Cortazar,1,2 Zoe A. Kibbelaar,3 Ilya G. Glezerman ,4 Ala Abudayyeh,5 Omar Mamlouk,5 Shveta S. Motwani,3,6 Naoka Murakami ,3 Sandra M. Herrmann,7 Sandhya Manohar ,7 Anushree C. Shirali,8 Abhijat Kitchlu,9 Shayan Shirazian,10 Amer Assal ,11 Anitha Vijayan,12 Amanda DeMauro Renaghan,13 David I. Ortiz-Melo,14 Sunil Rangarajan,15 A. Bilal Malik,16 Jonathan J. Hogan,17 Alex R. Dinh,17 Daniel Sanghoon Shin,18,19 Kristen A. Marrone,20 Zain Mithani,21 Douglas B. Johnson,22 Afrooz Hosseini,3 Deekchha Uprety,3 Shreyak Sharma ,3 Shruti Gupta ,3 Kerry L. Reynolds,23 Meghan E. Sise,1 and David E. Leaf 3 Due to the number of contributing authors, the affiliations are listed at the end of this article. ABSTRACT Background Despite increasing recognition of the importance of immune checkpoint inhibitor–associated AKI, data on this complication of immunotherapy are sparse. Methods We conducted a multicenter study of 138 patients with immune checkpoint inhibitor–associated AKI, defined as a $2-fold increase in serum creatinine or new dialysis requirement directly attributed to an immune checkpoint inhibitor. We also collected data on 276 control patients who received these drugs but didnotdevelopAKI. Results Lower baseline eGFR, proton pump inhibitor use, and combination immune checkpoint inhibitor therapy were each independently associated with an increased risk of immune checkpoint inhibitor– associated AKI. Median (interquartile range) time from immune checkpoint inhibitor initiation to AKI was 14 (6–37) weeks. Most patients had subnephrotic proteinuria, and approximately half had pyuria. Extrarenal immune-related adverse events occurred in 43% of patients; 69% were concurrently receiving a potential tubulointerstitial nephritis–causing medication. Tubulointerstitial nephritis was the dominant lesion in 93% of the 60 patients biopsied. Most patients (86%) were treated with steroids. Complete, partial, or no kidney recovery occurred in 40%, 45%, and 15% of patients, respectively. Con- comitant extrarenal immune-related adverse events were associated with worse renal prognosis, whereas concomitant tubulointerstitial nephritis–causing medications and treatment with steroids were each as- sociated with improved renal prognosis. Failure to achieve kidney recovery after immune checkpoint inhibitor–associated AKI was independently associated with higher mortality. Immune checkpoint inhib- itor rechallenge occurred in 22% of patients, of whom 23% developed recurrent associated AKI. Conclusions This multicenter study identifies insights into the risk factors, clinical features, histopathologic findings, and renal and overall outcomes in patients with immune checkpoint inhibitor–associated AKI. JASN 31: ccc–ccc, 2020. doi: https://doi.org/10.1681/ASN.2019070676 Received July 9, 2019. Accepted November 21, 2019. Immune checkpoint inhibitors (ICPis) are a novel class of immunotherapy that have revolutionized F.B.C. and Z.A.K. contributed equally to this work. 1,2 the treatment of a number of malignancies. Published online ahead of print. Publication date available at By targeting inhibitory receptors expressed on www.jasn.org. T lymphocytes, other immune cells, and tumor Correspondence: Dr. Frank B. Cortazar, New York Nephrology cells, these monoclonal antibodies enhance tumor- Vasculitis and Glomerular Center, 62 Hackett Boulevard, Albany, directed immune responses, and have been dem- NY 12209. Email: [email protected] onstrated to be highly effective in treating a broad Copyright © 2020 by the American Society of Nephrology JASN 31: ccc–ccc,2020 ISSN : 1046-6673/3102-ccc 1 CLINICAL RESEARCH www.jasn.org spectrum of malignancies.3 However, the increased antitumor Significance Statement activity achieved with these agents comes at the cost of a unique spectrum of autoimmune phenomena known as immune- Kidney toxicity from use of immune checkpoint inhibitors is being related adverse events (irAEs).4 The incidence of irAEs in recognized as an increasingly frequent complication of treatment. – patients receiving ICPis ranges from 60% to 85%, with the However, existing data on immune checkpoint inhibitor associated AKI have been limited to small, mostly single-center studies. In this skin, gastrointestinal tract, and liver being the most common multicenter study of 138 patients with immune checkpoint in- organs affected.1,5 Renal complications of ICPis, although less hibitor–associated AKI and 276 controls, the authors characterize common, are becoming increasingly recognized as the use of the clinical features of this complication and identify risk factors these agents continues to expand.6 associated with its development, clinicopathologic features, and The estimated incidence of immune checkpoint inhibitor– determinants of kidney recovery after an episode. Failure to achieve kidney recovery was associated with worse overall survival, and a 7 associated AKI (ICPi-AKI) ranges from 1.4% to 4.9%. Initial minority (23%) of patients who were retreated with immune reports described tubulointerstitial nephritis (TIN) as the checkpoint inhibitors had a recurrence of AKI. The study provides most common renal lesion caused by ICPis,7,8 although other insights into immune checkpoint inhibitor–associated AKI, although immune-mediated pathologies have also been reported, further study is needed to inform the care of affected patients. including various glomerulonephritides.9 Despite increasing recognition of the importance of patient characteristics as potential risk factors for ICPi-AKI. ICPi-AKI, our current understanding of ICPi-AKI is limited The temporal distribution of ICPi initiation was similar be- to case reports and small case series, the largest of which in- tween cases and controls (Supplemental Table 2). cluded 16 patients.9 We therefore conducted a multicenter, retrospective study comprising 138 patients with ICPi-AKI to Data Collection determine the independent risk factors for development of We collected detailed clinical data on all patients using a secure, ICPi-AKI, the clinical and pathologic features associated with standardized, electronic case report form (REDCap). Data col- ICPi-AKI, the key factors associated with kidney recovery after lected from each patient included demographics, comorbidi- an episode of ICPi-AKI, the risk of recurrent AKI with ICPi ties, use of concomitant potential TIN-causing medications, rechallenge, and the effect of ICPi-AKI on overall survival. longitudinal SCr and other laboratory values, kidney biopsy data, treatment data, data on ICPi rechallenge, and data on renal and overall outcomes. Additional details on clinical data METHODS collected are in the Supplemental Appendix 1. Overview Definitions of AKI Severity, Kidney Recovery, and We performed a multicenter, retrospective cohort study to assess Recurrent ICPi-AKI the clinical features and outcomes of ICPi-AKI. We contacted AKI severity was staged according to the Kidney Disease Im- nephrologists and oncologists at 26 major academic cancer proving Global Outcomes criteria.10 By definition, all cases centers across the United States and Canada to identify cases were stage 2 (doubling of SCr) or stage 3 (tripling of SCr or of ICPi-AKI. All protocols were approved by the Massachusetts need for RRT). We defined complete recovery of AKI as a re- General Hospital institutional review board, and by the insti- turn of SCr to ,0.35 mg/dl above the baseline value, whereas tutional review boards of all participating sites. we defined partial recovery as a return of SCr to .0.35 mg/dl but less than twice the baseline value, or liberation from RRT regard- Patients with ICPi-AKI less of the SCr value.7,11 Recurrent ICPi-AKI was defined as a Patients with ICPi-AKI were included if the AKI was attributed doubling of SCr or need for RRT after rechallenge with an ICPi. directly to the ICPi by the treating provider and the patient had at least a doubling of serum creatinine (SCr) or the require- Statistical Analyses ment for RRT. In total, 18 institutions (Supplemental Table 1) We performed the statistical analyses with Stata version 14 contributed 138 cases that met the above criteria for ICPi-AKI. (StataCorp., College Station, TX). Continuous and categorical data were compared using the Wilcoxon rank-sum and Fisher Control Patients exact tests, respectively. Univariate and multivariable logistic To identify risk factors for development of ICPi-AKI, we also regression were used to identify risk factors for incident collected data from 276 patients from contemporaneous ICPi-AKI, and, among those with ICPi-AKI, to identify factors registries at Dana Farber Cancer Institute (n=107) and associated with kidney recovery. We used the Kaplan–Meier Massachusetts General Hospital (n=169) who received ICPis method to estimate the cumulative death rate over time, and but did not develop AKI (case-to-control ratio of 1:2). All pa- Cox proportional hazards models to identify risk factors tients treated with ICPis were eligible to serve as controls except associated with increased mortality. Selection of covariates those who sustained an episode of AKI (defined as .50% in multivariable models was based on univariate associations increase in SCr). Random, rather than matched, selection of and biologic relevance. We used Schoenfeld residuals and controls was chosen to preserve the ability to investigate all ln-ln plots to verify that the proportional hazards